The 2022 State of Intelligent Virtual Assistants

Intelligent virtual assistants (IVAs) have great potential to streamline business processes, improve customer service, boost productivity, reduce expenses, and increase revenue. However, a number of pieces still need to be put into place before voice interfaces complete such tasks.

Suppliers continued to make progress in an ad hoc fashion, beefing up their product portfolios by acquisitions and forging more alliances. Developing standard interfaces is a to-do item that is expected to gain steam as we move through the new year.

Year in Review

“Speech platforms have come a long way in the last decade, and research in [natural language processing] and speech recognition and generation have evolved to provide generic speech platforms to use for several languages for applications,” explains Balaji Raghavan, chief technology officer of Uniphore. “This enables building quick prototypes and getting a product available to users quickly in the cloud. Despite this enablement, there are still a bunch of challenges to deliver this technology in applications as a [software-as-a-service] product to enterprises.”

Vendors tried to fill the various voids by acquiring needed technology from other voice suppliers. A year ago, Nuance Communications bought Saykara, a startup developing a mobile AI assistant that automates clinical documentation generation for physicians. The acquisition underscores Nuance’s ongoing expansion into conversational artificial intelligence (AI) and ambient clinical intelligence (ACI). The solution was built to reduce clinician burnout, enhance patient experiences, and improve overall health system financial integrity.

Two months later, Microsoft shook up the market by paying $19.7 billion to acquire Nuance. “Microsoft and Nuance both focused more on enterprise voice applications than consumer solutions, so there is synergy in their target markets,” notes Deborah Dahl, principal at Conversational Technologies.

In August, NICE purchased GoMoxie, which makes digital tools that anticipate customer needs and guides them through their online journeys by presenting agents with relevant snippets of information. Such features have the potential to increase customer use of conversational self-service interactions.

In September, Verint completed the acquisition of Conversocial, which focuses on conversational customer experience delivered over various messaging channels, such as Apple Business Chat, Facebook Messenger, Twitter, and WhatsApp.

LivePerson had a busy year rounding out its product line. In July, the vendor bought German conversational AI company e-bot7. The acquisition propped up LivePerson’s self-service capabilities with the ability to automate customer interactions by suggesting relevant topics and generating replies based on context-related information and natural language understanding. Then in October, the vendor made two more purchases: VoiceBase, a supplier of real-time speech recognition and conversational analytics; and Tenfold, a customer engagement platform integrated with leading CRM solutions.

Also in October, Uniphore purchased Jacada. Uniphore makes conversational AI for sentiment, intent, and emotion analysis. Jacada specializes in workflow and desktop automation. The end goal is to assist agents relying on Uniphore’s U-Assist, so as they process real-time calls, the system presents them with next-best-action advice based on customer intent, conversation content, sentiment, and emotion.

Keeping with the theme of a busy October, Vonage added conversational commerce to its product portfolio with the acquisition of Jumper.ai. The conversational commerce solution attempts to capitalize on the growing convergence of shopping and conversations on social platforms, such as Facebook Messenger, WhatsApp, and Instagram, by using chat features to create seamless shopping experiences.

Partnerships were also plentiful as IVA suppliers extended their reach by adding support for more computing platforms. In November, NICE worked with Google Cloud to integrate its cloud-based, AI-powered CXone customer experience platform with Google Cloud Contact Center Artificial Intelligence (CCAI), a group of APIs that bring Google AI capabilities to contact center use cases. The new combination provides businesses with sophisticated and efficient ways to engage with customers across digital and voice touchpoints.

LivePerson also joined forces with Google Cloud. The former’s Conversational Cloud now includes Google Cloud CCAI’s Agent Assist, so agents have real-time recommendations for replies to customer messages, as well as suggested text to complete sentences at their fingertips.

IVA technology has been discussed for more than a decade, but increasingly, the talk is being transformed into action. Solutions like Apple’s Siri, Amazon’s Alexa, and Google Assistant are now available to nearly half the people in the world. Billions of people have smartphones with assistants and millions have smart speakers. The voice solutions gained traction in the consumer market and perform tasks like turning on the heat in an apartment and providing the latest score for the home hockey team.

Enterprises have also been taking notice. “Somewhere in the last year or so, corporations recognized that what people say over the phone is potentially a major asset,” explains Dan Miller, founder and lead analyst of Opus Research. “The combination of call recording, contact center, IVR, and voice bots best signals a person’s intent, so they are putting tools in place to capture and analyze that information so they understand their customers’ intentions in real time.”

A growing number of enterprises are deploying IVAs. Almost three-quarters (74 percent) agreed that controlling their brand identity and user experiences are the greatest drivers of value, according to an Opus Research study.

Customer service has been one area where the technology fits well, and the evolution from interactive voice response solutions to IVAs seems to be a natural transition. In fact, more than two-thirds (67 percent) of companies already employ IVAs to interact with customers, clients, or prospects, Opus found.

Consequently, investments in voice technology have grown as the market evolves. BCC Research expects the global IVA market to grow from $2.7 billion in 2021 to $7.6 billion by 2026, a compound annual growth rate of 22.8 percent.

As the growth occurs, the market has been shifting from a horizontal to a vertical focus. “Some of the challenges that speech vendors face involve being able to provide the ability to customize with domain-specific entities and rules,” says Uniphore’s Raghavan.

In November, Conversica delivered an account-based marketing IVA. The solution automates personalized outreach across email, SMS, and web chat. By examining items such as firmographic, buyers’ persona, sales stage, and intent data, the tool determines customer’s intent, identifies the next best action, presents it to the agent, and tries to prod potential customers toward a purchase.

Healthcare has been an area of emphasis. In May, Athena Health integrated Nuance’s cloud-based Dragon Medical IVA into athenaOne’s EHR and mobile applications. By leveraging Nuance’s conversational AI platform, athenaOne Dictation users take advantage of integrated, voice-driven capabilities to reduce documentation time and streamline patient care.

Finance is another space where IVAs continue to gain traction. In August ConsenSys partnered with LivePerson to provide conversational AI support for its blockchain and crypto communities, starting with users of leading crypto wallet MetaMask.

More work in this area is expected in 2022. Integrating these features with other business processes could help companies remove friction, such as filling out paperwork, from existing business processes. “Companies want to use technology to create end-to-end, seamless transactions and interactions,” explained Joshua Greenbaum, an industry analyst and consultant at Enterprise Applications Consulting (EAC).

Voice is becoming one interface in these transactions, and vendors are tying it into various workflow solutions. In December, NICE introduced AI-powered capabilities that enable organizations to maximize the benefits of robotic process automation (RPA). Included in version 7.6, NICE RPA’s new capabilities include document digitization, ROI-based recommendation of ideal processes to automate, and a complimentary resource center with ready-made low-code/no-code resources for sharing. In addition to reducing process analysis time and automating manual tasks, the innovative new capabilities also help organizations boost ROI and maximize the value of automation projects for the business.

A Look Ahead

While the market has made progress, much more work needs to be done. Developers can now create specific add-on applications accessible through the generic assistants, using tools like the Alexa Skills Kit or Google DialogFlow.

Many features have been developed, but these systems often have problems running on more than one platform. Alexa Skills work only on Amazon devices and DialogFlow applications are relegated to Google. This limitation means that if an organization wants to use voice assistants for customer support, it has to develop and, perhaps even more importantly, maintain several versions of the software, increasing housekeeping requirements, slowing new deployments, and boosting costs.

Help might be on the way. The Stanford Open Virtual Assistant Lab Project developed Genie, an open-source IVA. The goal is to mimic World Wide Web browsers and empower users so they work with voice systems in any way they desire. The group is outlining standards so IVA solutions scale up cost-effectively. Its Pretrained Agent Generator produces transactional dialogue agents from just database schemas, API signatures, and a few samples of natural language utterances that others can use in their voice applications.

The IVA market has gradually been taking shape. In 2021, the technology gained significant traction, and use became common in areas like contact centers. Now in 2022, the move to more vertical applications and the emergence of standard interfaces has the potential to continue the momentum. 

Paul Korzeniowski is a freelance writer who specializes in technology issues. He has been covering speech technology issues for more than two decades, is based in Sudbury, Mass., and can be reached at paulkorzen@aol.com or on Twitter @PaulKorzeniowski.